click below
click below
Normal Size Small Size show me how
WEEK 5:
Comparing people: case control studies:
| Question | Answer |
|---|---|
| case control study | compares people with and without disease to explore factors that contribute to difference between groups (identifies people who are already ill ,cases, and a group who are not , controls) while looking backwards in time |
| design of case control study | population split into cases (people with disease) and controls (people without disease) and both are split into exposed and non exposed |
| strengths of case control study | no need to follow people up over time, efficient for rare diseases, can study multiple exposures, and potentially very suitable when you want quick results (eg infectious disease outbreak where you start with those ill and look for risk factors) |
| limitations of case control study | only look how exposures are associated with one outcome, cannot make data on incidence (no RR), prone to bias, hard if record kept is inadequate/unreliable, selection of controls can be difficult, difficult to know if exposure comes after disease |
| case control studies are inefficient for | rare exposures |
| causation as a limitation in case control studies | causation cannot formally be established |
| why cant we generate data on incidence/ calculate a risk ration RR in case control studies | in case control studies we start with the outcome so if we select people with disease and people without disease we cannot calculate risk of getting lung cancer as we already selected who has it already and not |
| incidence risk | measure of how many new cases (outcome) arise in a population that is susceptible over a period of time |
| risk ratio RR | comparing chances of that outcome occurring in an exposed and unexposed group |
| what ratio can you use in case control studies | odds ratio OR (measures association between factor and disease) |
| stages of case control study (5) | Define a case group Define a control group Sample cases and controls Collect exposure data Analyse and interpret data |
| how to define case group | requires precise case definition and eligibility criteria and somewhere to source cases |
| incident cases | new cases of disease eg first admissions |
| prevalent cases | people living with diagnosed disease (though they may have changed their lifestyle since diagnosis) |
| best controls when defining control group | best controls are drawn from same population as cases |
| hospital controls | can be difficult to identify from safe clinic and watch out for atypically high or low prevalence of exposure |
| effect of more controls | can increase precisions of the odds ratio (control group and case group do not have to be the same number) |
| selection bias | cases and controls are not representative of people with and without the disease |
| sample cases and controls include | types of sampling (Random, systematic and matched etc) |
| random sampling identified from | list of all potential participants |
| systematic sampling | sequential selection using every n'th person |
| matched sampling | control paired with case (eg age, sex, hospital) |
| collecting exposure data includes types of bias like | recall bias, response bias, and interviewer bias - so better if there are objective measures |
| recall bias | selective memory of issues related to exposure and or outcome (eg people may remember their exposure in greater detail because they are ill) |
| response bias | if participants respond in the same way they think are expected and or wanted to respond (social desirability) |
| interviewer bias | if interviewer knows whether participants are a case of control and introduces own bias |
| number for null hypothesis | 1 |